The present invention relates to a gradation image processing apparatus and a method for expressing a halftone-including image in a facsimile or the like with binary black-and-white pixels.
In an apparatus such as a facsimile for scanning and sampling an original to read out its image per pixel unit and copying the image thus read out, the purpose of the apparatus is not to achieve a faithful reproduction of the image. An improvement of the image is attempted by processing the image signal to emphasize or extract specific data of the original image for eliminating the noise component included in the original image or for restoring the sharpness of the image.
In a conventional image processing method which eliminates noise components in an original image, the noise is smoothed by first detecting the densities of specific pixels adjacent to a subject pixel, and taking the accumulated density frequency distribution per sampling level to process an average value as the density level of the subject pixel. Although such an image processing method is effective to smooth the reproduced image by eliminating of granular noise, disadvantages exist that include a loss of the image signal steepness in a portion of the original image where the density change is great, i.e. a loss of the image sharpness which eventually renders the image indistinct. Furthermore, it is impossible to eliminate the noise that is derived from the soiled surface of the original.
According to another conventional image processing method, the minimal level of specific pixels adjacent to a subject pixel is subtracted from the density level of the subject pixel. The value thus obtained is regarded as the true density level of the latter pixel. However, in an image processing method, the densitylevel variation range is reduced which consequently brings about a lower contrast of the image as a whole. Another unavoidable drawback is that the granular noise cannot substantially be eliminated.